Sergey
@sergeycyw.bsky.social
I talk about growth stock investing and fundamental analysis with a long-term mindset. I provide earnings reviews and key news updates. Not investment advice.
SergeyCYW.substack.com
SergeyCYW.substack.com
The pattern is clear. Google is building scale across cloud, models, and real-world AI deployment. The open question is whether capacity and power can expand fast enough to keep up with demand.
December 19, 2025 at 4:04 PM
The pattern is clear. Google is building scale across cloud, models, and real-world AI deployment. The open question is whether capacity and power can expand fast enough to keep up with demand.
New enterprise and consumer offerings are increasing utilization, while profitability remains a longer-term objective given infrastructure and regulatory costs.
December 19, 2025 at 4:04 PM
New enterprise and consumer offerings are increasing utilization, while profitability remains a longer-term objective given infrastructure and regulatory costs.
Waymo adds a long-duration option on autonomy. Commercial expansion is underway across multiple U.S. cities, with international launches planned for London and Tokyo ahead of 2026.
December 19, 2025 at 4:04 PM
Waymo adds a long-duration option on autonomy. Commercial expansion is underway across multiple U.S. cities, with international launches planned for London and Tokyo ahead of 2026.
Developers are leaning in, with 13M+ building on Google’s models and API traffic running at 7B tokens per minute. Veo 3 has already generated 230M+ videos, now embedded into YouTube and marketing workflows. Gemini Enterprise surpassed 2M paid seats across 700 companies.
December 19, 2025 at 4:04 PM
Developers are leaning in, with 13M+ building on Google’s models and API traffic running at 7B tokens per minute. Veo 3 has already generated 230M+ videos, now embedded into YouTube and marketing workflows. Gemini Enterprise surpassed 2M paid seats across 700 companies.
Demand is strong enough that capacity, not customers, has become the primary constraint.
Usage is scaling at unprecedented speed. Gemini reached 650M monthly active users, with query volume up 3x since Q2. Across Google’s platforms, token processing hit 1.3 quadrillion per month, up 20x YoY.
Usage is scaling at unprecedented speed. Gemini reached 650M monthly active users, with query volume up 3x since Q2. Across Google’s platforms, token processing hit 1.3 quadrillion per month, up 20x YoY.
December 19, 2025 at 4:04 PM
Demand is strong enough that capacity, not customers, has become the primary constraint.
Usage is scaling at unprecedented speed. Gemini reached 650M monthly active users, with query volume up 3x since Q2. Across Google’s platforms, token processing hit 1.3 quadrillion per month, up 20x YoY.
Usage is scaling at unprecedented speed. Gemini reached 650M monthly active users, with query volume up 3x since Q2. Across Google’s platforms, token processing hit 1.3 quadrillion per month, up 20x YoY.
Backlog climbed to $155B, up 82% YoY, driven by long-duration AI contracts. Billion-dollar deals signed this year already exceeded the prior two years combined. More than 70% of Cloud customers now use Google AI products, with nearly 150 enterprises processing ~1T tokens each annually.
December 19, 2025 at 4:04 PM
Backlog climbed to $155B, up 82% YoY, driven by long-duration AI contracts. Billion-dollar deals signed this year already exceeded the prior two years combined. More than 70% of Cloud customers now use Google AI products, with nearly 150 enterprises processing ~1T tokens each annually.
Companies controlling compute, memory, cloud infrastructure, and orchestration layers are capturing outsized value, while semiconductor leaders benefit from sustained demand visibility rather than short cycles.
December 19, 2025 at 3:00 PM
Companies controlling compute, memory, cloud infrastructure, and orchestration layers are capturing outsized value, while semiconductor leaders benefit from sustained demand visibility rather than short cycles.
The AI buildout has entered a phase defined less by experimentation and more by industrial-scale execution.
December 19, 2025 at 3:00 PM
The AI buildout has entered a phase defined less by experimentation and more by industrial-scale execution.
By late 2024, Tesla deployed 50,000 GPUs in Texas and added 16,000 H200 GPUs in Q2 2025, bringing the Cortex cluster to ~67,000 H100 equivalents. Capex exceeded $11B for the year, reflecting a clear shift toward NVIDIA-based AI training at scale.
December 19, 2025 at 3:00 PM
By late 2024, Tesla deployed 50,000 GPUs in Texas and added 16,000 H200 GPUs in Q2 2025, bringing the Cortex cluster to ~67,000 H100 equivalents. Capex exceeded $11B for the year, reflecting a clear shift toward NVIDIA-based AI training at scale.
FSD generated $326M in quarterly revenue, supported by new feature releases. AI training compute expanded more than 75% year over year, with over 2B miles driven cumulatively on FSD Supervised.
December 19, 2025 at 3:00 PM
FSD generated $326M in quarterly revenue, supported by new feature releases. AI training compute expanded more than 75% year over year, with over 2B miles driven cumulatively on FSD Supervised.
$TSLA Tesla
Tesla continues to scale AI compute to support Full Self-Driving development, even after exiting the Dojo custom supercomputer effort. Q3 2024 revenue reached $25.2B, up 8% YoY, while net income rose to $2.17B and operating cash flow hit a record $6.3B.
Tesla continues to scale AI compute to support Full Self-Driving development, even after exiting the Dojo custom supercomputer effort. Q3 2024 revenue reached $25.2B, up 8% YoY, while net income rose to $2.17B and operating cash flow hit a record $6.3B.
December 19, 2025 at 3:00 PM
$TSLA Tesla
Tesla continues to scale AI compute to support Full Self-Driving development, even after exiting the Dojo custom supercomputer effort. Q3 2024 revenue reached $25.2B, up 8% YoY, while net income rose to $2.17B and operating cash flow hit a record $6.3B.
Tesla continues to scale AI compute to support Full Self-Driving development, even after exiting the Dojo custom supercomputer effort. Q3 2024 revenue reached $25.2B, up 8% YoY, while net income rose to $2.17B and operating cash flow hit a record $6.3B.
Capital expenditures totaled $13B in Q3, primarily for AI infrastructure, with further increases expected in 2025 as demand scales.
December 19, 2025 at 3:00 PM
Capital expenditures totaled $13B in Q3, primarily for AI infrastructure, with further increases expected in 2025 as demand scales.
AI-driven products built on Gemini expanded revenue over 200% YoY, with API usage rising 14x in six months. Over 70% of Cloud customers now use AI services, and billion-dollar deal signings exceeded the prior two years combined.
December 19, 2025 at 3:00 PM
AI-driven products built on Gemini expanded revenue over 200% YoY, with API usage rising 14x in six months. Over 70% of Cloud customers now use AI services, and billion-dollar deal signings exceeded the prior two years combined.
$GOOGL Alphabet
Alphabet continues monetizing AI through Google Cloud and generative AI services. Q3 2024 revenue reached $88.3B, up 15% YoY, while Google Cloud revenue grew 35% to $11.4B with 17% operating margin.
Alphabet continues monetizing AI through Google Cloud and generative AI services. Q3 2024 revenue reached $88.3B, up 15% YoY, while Google Cloud revenue grew 35% to $11.4B with 17% operating margin.
December 19, 2025 at 3:00 PM
$GOOGL Alphabet
Alphabet continues monetizing AI through Google Cloud and generative AI services. Q3 2024 revenue reached $88.3B, up 15% YoY, while Google Cloud revenue grew 35% to $11.4B with 17% operating margin.
Alphabet continues monetizing AI through Google Cloud and generative AI services. Q3 2024 revenue reached $88.3B, up 15% YoY, while Google Cloud revenue grew 35% to $11.4B with 17% operating margin.
3nm reached full capacity in 2024, delivering 15% performance gains and 30% power savings over 5nm. The 2nm node, scheduled for late 2025 production, already exceeds 60% yield, offering further efficiency gains. Advanced packaging technologies such as CoWoS remain critical enablers of AI systems.
December 19, 2025 at 3:00 PM
3nm reached full capacity in 2024, delivering 15% performance gains and 30% power savings over 5nm. The 2nm node, scheduled for late 2025 production, already exceeds 60% yield, offering further efficiency gains. Advanced packaging technologies such as CoWoS remain critical enablers of AI systems.
TSMC
TSMC remains the backbone of the AI semiconductor supply chain. High-performance computing accounted for 57% of revenue in Q3 2025, with advanced nodes at 7nm and below representing 74% of wafer revenue. Full-year 2025 growth was revised to the mid-30% range, driven by AI demand.
TSMC remains the backbone of the AI semiconductor supply chain. High-performance computing accounted for 57% of revenue in Q3 2025, with advanced nodes at 7nm and below representing 74% of wafer revenue. Full-year 2025 growth was revised to the mid-30% range, driven by AI demand.
December 19, 2025 at 3:00 PM
TSMC
TSMC remains the backbone of the AI semiconductor supply chain. High-performance computing accounted for 57% of revenue in Q3 2025, with advanced nodes at 7nm and below representing 74% of wafer revenue. Full-year 2025 growth was revised to the mid-30% range, driven by AI demand.
TSMC remains the backbone of the AI semiconductor supply chain. High-performance computing accounted for 57% of revenue in Q3 2025, with advanced nodes at 7nm and below representing 74% of wafer revenue. Full-year 2025 growth was revised to the mid-30% range, driven by AI demand.
Open-source strategy prioritizes ecosystem scale and standardization rather than near-term monetization.
December 19, 2025 at 3:00 PM
Open-source strategy prioritizes ecosystem scale and standardization rather than near-term monetization.
Capital expenditures hit $9.2B in Q3, with full-year $38–40B guidance, rising sharply to $70–72B in 2025 to fund AI data centers and compute. Llama surpassed 650M downloads, with enterprise adoption accelerating.
December 19, 2025 at 3:00 PM
Capital expenditures hit $9.2B in Q3, with full-year $38–40B guidance, rising sharply to $70–72B in 2025 to fund AI data centers and compute. Llama surpassed 650M downloads, with enterprise adoption accelerating.
$META Meta Platforms
Meta continues to pair AI monetization with aggressive infrastructure investment. Q3 2024 revenue reached $40.6B, up 19% YoY, while net income climbed 35% to $15.7B. Operating margin expanded to 43% as AI-driven ad ranking improved pricing and engagement.
Meta continues to pair AI monetization with aggressive infrastructure investment. Q3 2024 revenue reached $40.6B, up 19% YoY, while net income climbed 35% to $15.7B. Operating margin expanded to 43% as AI-driven ad ranking improved pricing and engagement.
December 19, 2025 at 3:00 PM
$META Meta Platforms
Meta continues to pair AI monetization with aggressive infrastructure investment. Q3 2024 revenue reached $40.6B, up 19% YoY, while net income climbed 35% to $15.7B. Operating margin expanded to 43% as AI-driven ad ranking improved pricing and engagement.
Meta continues to pair AI monetization with aggressive infrastructure investment. Q3 2024 revenue reached $40.6B, up 19% YoY, while net income climbed 35% to $15.7B. Operating margin expanded to 43% as AI-driven ad ranking improved pricing and engagement.
Demand remains strong, with AI accelerator revenue expected to double in 2025, reinforcing NVIDIA’s structural leadership across training and inference.
December 19, 2025 at 3:00 PM
Demand remains strong, with AI accelerator revenue expected to double in 2025, reinforcing NVIDIA’s structural leadership across training and inference.
The Blackwell architecture entered volume production with 208B transistors, delivering 20 petaflops FP4 per GPU and up to 30x inference performance versus Hopper at dramatically higher energy efficiency.
December 19, 2025 at 3:00 PM
The Blackwell architecture entered volume production with 208B transistors, delivering 20 petaflops FP4 per GPU and up to 30x inference performance versus Hopper at dramatically higher energy efficiency.
$NVDA NVIDIA
NVIDIA continues to dominate AI compute infrastructure. Fiscal Q3 2026 revenue reached $57B, up 62% YoY, with data center revenue of $51.2B, rising 66% YoY. Gross margin remained resilient at ~73%, despite massive volume ramps. Net income totaled $31.9B, with EPS of $1.30.
NVIDIA continues to dominate AI compute infrastructure. Fiscal Q3 2026 revenue reached $57B, up 62% YoY, with data center revenue of $51.2B, rising 66% YoY. Gross margin remained resilient at ~73%, despite massive volume ramps. Net income totaled $31.9B, with EPS of $1.30.
December 19, 2025 at 3:00 PM
$NVDA NVIDIA
NVIDIA continues to dominate AI compute infrastructure. Fiscal Q3 2026 revenue reached $57B, up 62% YoY, with data center revenue of $51.2B, rising 66% YoY. Gross margin remained resilient at ~73%, despite massive volume ramps. Net income totaled $31.9B, with EPS of $1.30.
NVIDIA continues to dominate AI compute infrastructure. Fiscal Q3 2026 revenue reached $57B, up 62% YoY, with data center revenue of $51.2B, rising 66% YoY. Gross margin remained resilient at ~73%, despite massive volume ramps. Net income totaled $31.9B, with EPS of $1.30.
The Artificial Intelligence Platform continues converting pilots into production deployments, pushing Palantir’s Rule of 40 score to 81 in Q4.
December 19, 2025 at 3:00 PM
The Artificial Intelligence Platform continues converting pilots into production deployments, pushing Palantir’s Rule of 40 score to 81 in Q4.
Adjusted operating margin reached a record 45%, highlighting strong operating leverage. Customer count increased 43% to 711, with 120% net dollar retention. Adjusted free cash flow totaled $1.25B, a 44% margin.
December 19, 2025 at 3:00 PM
Adjusted operating margin reached a record 45%, highlighting strong operating leverage. Customer count increased 43% to 711, with 120% net dollar retention. Adjusted free cash flow totaled $1.25B, a 44% margin.